Scope probes: techniques and builds

[Radio Wrangler]

Bufo Bill posted a request for HP1121A probes. Craig and myself posted that he wouldn't actually need them and that everyone uses alternatives except for a few specialised cases where the FET probes really are needed.

Probing techniques are likely to be of general interest to anyone with an oscilloscope, a spectrum analyser or a network analyser, so we've set up this thread and moved things here.

The forum software is timestamp based, so Bill's intro comes after the two moved posts...

On with the moved posts:


You don't actually need them.

I have some and they get used once a blue moon.

They are not only very expensive, they are also incredibly easily blown up (the thin film microcircuit in the probe end) consequently, any that are actually known to work are extremely expensive.

A lump of 50 ohm coax with a 5k resistor on the end and a pin soldered on is quite useful.

A spectrum analyser has a crazy amount of sensitivity compared to scopes and voltmeters. So you use it to allow the use of very lossy probes.

The FET probes are needed, but only when the easier cheats run out.

David

[Craig Sawyers]

I have the very similar Tektronix FET probes (which I bought pretty cheap NOS before prices got stupid). And like David they almost never get used.

A DIY Z0 probe as per David's suggestion is a good idea.

This guy https://emcesd.com/ has plans for a compensated 1GHz DIY Z0 probe. The attenuation can be changed by the value of the tip resistor.

https://emcesd.com/1ghzprob.htm

In fact his entire site is worth poking around. His bread and butter is RFI compatibility and running seminars. But the practical tips are excellent. Like using a paperclip as a magnetic loop to sniff out high frequency circulating currents.

Craig

[Bufo Bill]

David ( @Radio Wrangler ) suggested I start a thread here to learn about Oscilloscope and Spectrum Analyser probes and techniques in general. He may move some posts here from another thread.
David and @Craig Sawyers suggested I may be able to build my own probe suitable for a HP 8558B Spectrum Analyser; I am very interested in this idea and would love to give it a go. Hopefully one of them will chime in soon.

[Pamphonica]

I rolled my own proble for my spectrum analyser. The main problem is of course that the SA presents a 50 ohm load so will load low-level high impedance signals.
However, a simple 500 ohm probe can be constructed that gives an order of magnitude less circuit loading.
Just re-make an old scope lead, with just a 450 ohm resistor very near the tip. That gives 10:1 with the SA's input impedance. The BNC on the end can either be replaced with an N-type, or left on and used with an adapter.
A bit rough and ready, but a cheap and quick probe I quite often use.
Watch out for some tek /10 probes where the series resistance can be distributed between tip and connector! That fooled me for a bit.
Just remember to add 10dB to the readings on screen.
-Jeremy

[regenfreak]

I paid £9 on ebay for my RF active probe. I rarely used it because i want both my hands free in alignment. I will probably replace the needle and ground wire by micro clips.

To address the loading effect of the probe at high frequency, I add an resistor solder inside a croc slip keeping the lead short. The value of the resistor depends on the frequency of the measurement and the response curve of your probe. I use 100k to 1M for 10.7MHz probing.

[Pamphonica]

Originally Posted by regenfreak

I paid £9 on ebay for my RF active probe. I rarely used it because i want both my hands free in alignment. I will probably replace the needle and ground wire by micro clips.

That's an interesting device, which comes in various versions. Cheap enough to buy to play with! I might get one.

My fairly crude 500 ohm probe only has one advantage - it is passive so needs no power source. Otherwise it's fairly basic. I suppose you could make it a 5K ohm probe if you wanted (/100 or 20dB loss) using a 4.95K resistor (or resistors) on the input. Horses for courses!

An active probe would give the lowest loading on the circuit.

-Jeremy

[Skywave]

On the subject of oscilloscopes and probes, two points that sometimes get over-looked:

1. If your 'scope claims a 300 MHz bandwidth (say) that does not mean that you can accurately measure signals up to 300 MHz. That stated 'bandwidth' is for - 3dB. Your 'scope will probably have a flat response to something like 100 MHz, not 300 MHz.

2. Every 'scope probe has a 'bandwidth' too. So if you're using your 'scope with a probe, the overall 'flat bandwidth' for measurement will be the combined effect of the b/w of the 'scope and the b/w of the probe. Consequently, that 'flat b/w' will be less than the b/w of the 'scope or the probe separately.

Al.

[Radio Wrangler]

A 500 Ohm resistive probe can outperform an FET active probe at higher frequencies, it can be made to have somewhat lower stray capacitance than an FET's gate. All sorts of bootstrap arrangements can be used to reduce the FET capacitance, but they eventually run out of steam. 2pf at 100MHz has a reactance of 795 Ohms. 79.5 Ohms at a gigahertz.

OK, so someone says they're not going over X MHz, but with a spectrum analyser, you'll want to see the harmonic amplitudes, you'll want to see the frequencies at which spurious oscillation can occur, even if just to see that there aren't any. With a network analyser, you'll want to go high so you can follow the skirts of a filter to where they've rolled off a lot, to check that unplanned spurious resonances don't bring it back up again.

Your choice of probe isn't so much made to suit the instrument you're using as to suit the circuit you're prodding.

David

[Bufo Bill]

I may build a couple of the different designs people have mentioned, just to get my head around the practical opportunities each one can offer.
Thanks to all who have posted so far.
Cheers from Bill.

[woodchips]

I would be surprised if a 300MHz scope was only flat to 100MHz. The bandwidth is normally specified at the front panel connector, so will be 300MHz -3dBm. Also because analogue scopes are peaked then the drop off rate of the amplitude will be much faster than 6dB octave. With digital scopes the drop off rate will be like a stone as the half sampling rate is reached.

As been said, miniature coax with a 1/8W resistor soldered on the end works well to multiple GHz.

If really want bandwidth then use a sampling scope. Because their input is wide open then they do have a 6dB octave drop off rate. I have used the Tek S2 plugin, rated 4.7GHz, at signals over 10GHz. Have to allow for the extra loss but the signal is clearly visible, just smaller. The Tek 7S11/7T11 are amazing units, will actually trigger at 12GHz, the HP sampling stuff always gave up at 1GHz.

[Radio Wrangler]

Classic analogue scopes have several amplifier stages, each with frequency response correction. The manufacturers try to be sure that instruments comfortably meet the 2dB bandwidth spec, so yo'll find them good up to maybe three quarters of the rated bandwidth, maybe -2dB at it and then starting to fall fast as several stages roll off together.

Another roll-off factor is the length of the Y plates and the velocity of the electrons. Too high a frequency and the averaging effect of the net deflection reduces sensitivity. It's just like the effect of tape head gap and tape speed.

This is where distributed plates built as a delay line to match the electron speed come in.

David

[G0HZU_JMR]

I have a boxful of different RF probe types here including the classic passive Zo probes already described. However, for LW/MW/SW stuff I would usually reach for my old Marconi TK2374 200MHz FET probe. This probe is at its best with the x10 attenuator tip attached as the input capacitance is then only about 2.5pF and the parallel resistance is very high. This makes it very useful for probing high impedance circuits either in the RF or IF or even the AF stages as it is quite flat down to about 500Hz (-3dB at 50Hz).

Care is needed when using it without the x10 or x100 attenuator tips because the probe has a 100k resistor directly shunted across the tip when set to x0 attenuation (no attenuator tip fitted). This limits the input impedance and also can upset the circuit under test as it can affect the biasing of an amplifier.

I modified the dc power feed of my TK2374 so it can be used with any spectrum analyser. Otherwise it will only work with the old 110MHz Marconi TF2370 analyser as the dc power feed has a custom connector that mates with the front panel of the TF2370. It might also work (as in be able to be dc powered) with the Mi2382 analyser as well but I can't remember.

It can cope with 40V peak RF (as in damage level) with the x10 head attached and it doesn't matter if you blow it up because it uses a basic JFET buffer amp in the probe tip. It is really easy to take apart and repair. Everyone with a network/spectrum analyser should have one! I even use it with a scope sometimes to measure the resonator voltage swing on a HF oscillator.

[Radio Wrangler]

Sounds daft, it's definitely not flat, but you can do a lot screwing a normal scope probe onto the front of a spectrum analyser. Yes it's a 50 Ohm input when the probe is for 1 meg, but it works if you're not bothered about amplitude accuracy and just want a rough sniff of RF

Try it!

David

[Craig Sawyers]

Originally Posted by Radio Wrangler

Classic analogue scopes have several amplifier stages, each with frequency response correction. The manufacturers try to be sure that instruments comfortably meet the 2dB bandwidth spec, so yo'll find them good up to maybe three quarters of the rated bandwidth, maybe -2dB at it and then starting to fall fast as several stages roll off together.

Another roll-off factor is the length of the Y plates and the velocity of the electrons. Too high a frequency and the averaging effect of the net deflection reduces sensitivity. It's just like the effect of tape head gap and tape speed.

This is where distributed plates built as a delay line to match the electron speed come in.

David

Tek took that to the extreme with the CRT in the 7104. With the right plugins it will do DC-1GHz real time. And before the phosphor it has an electron multiplier microchannel plate. That has such a high gain that it is possible to see a single shot at 200ps per division in bright daylight.

At the deflection plates of that superb CRT have a bandwidth of >2GHZ.

But it is dangerous in unskilled hands. Turn the brightness up and you can easily damage the microchannel plate. In most examples that have been treated kindly the screen readout has reduced sensitivity, because it is has been running all the time. Unless you are knowledgeable and use either the readout off detente, run it gated (so it only appears when the trace is triggered), trigger it externally via a rear panel BNC, or push the "manual" button when you want one one readout scan.

Mine gets used on high days and holidays. When a scope of no lower capability will do.

Craig

[Dickie]

I wouldn't suggest anybody buys one of these, but we used them at my last employer and show how far a "simple" resistive probe can go.

http://www.auburntec.com/

[G0HZU_JMR]

With the passive Zo probes you do have to be mindful of mismatch uncertainty if you want to use it over a very wide bandwidth of >1GHz. If you want to start using a -20dB Zo probe with the series 450R resistor for example, you have to use very good quality 50R RF coaxial cable and the termination device (analyser?) at the far end of the cable needs to have low input VSWR across the relevant bandwidth.

A decent lab spectrum analyser with the default 10dB attenuation is usually adequate in terms of input VSWR up inn the GHz region but a 50R terminated scope won't be and I'd guess the latest generation of low cost analysers from China won't have low VSWR unless an external attenuator is fitted. It wouldn't be unusual to see +/-1.5dB of uncertainty in a typical setup for example.

If you try using a longish run of skinny coax up at many GHz with a -20dB probe like this it could easily result in +/- 3dB of uncertainty at several GHz due to the limitations of the coax alone.

[G0HZU_JMR]

Note that by skinny coax I mean something cheap and cheerful like RG174 sourced from ebay. I just clicked on the Auburntec probe link and it uses what looks like RG316 cable. This cable is skinny too but it is a decent UHF RF cable that would perform a lot better across several GHz.

[regenfreak]

his guy https://emcesd.com/ has plans for a compensated 1GHz DIY Z0 probe. The attenuation can be changed by the value of the tip resistor.

https://emcesd.com/1ghzprob.htm

In fact his entire site is worth poking around. His bread and butter is RFI compatibility and running seminars. But the practical tips are excellent. Like using a paperclip as a magnetic loop to sniff out high frequency circulating currents.

Craig

The use of copper tape as EMI in DIY probes/devices has been problematic for me due to the thin layer of glue causing poor conductivity, this guy's simple solution is here:


https://emcesd.com/tt2005/tt120105.htm

[peter_scott]

Originally Posted by G0HZU_JMR

I have a boxful of different RF probe types here including the classic passive Zo probes already described. However, for LW/MW/SW stuff I would usually reach for my old Marconi TK2374 200MHz FET probe.

I built myself one of these to work with my TF2370. The plug for powering was a little problematic but I found that I could modify an HP probe power plug to fit. It's quite a nice little accessory.

Peter.

[Skywave]

Originally Posted by Radio Wrangler

Classic analogue scopes: the manufacturers try to be sure that instruments comfortably meet the 2dB bandwidth spec.

"2dB bandwidth spec."? I've only ever seen 'scope B/W specs. stated at -3dB.

Al.